Mexico City Air Quality Research Initiative F. Guzman, G.E. Streit

Mexico City air quality research initiative

F. Guzman, G.E. Streit* Gerencia de Energeticos Alternos y Quimica Ambiental,

Institute Mexicano del Petroleo, Eje Central 152, 07730 Mexico, D.F., Mexico

*Los Alamos National Laboratory, Group A-4, Mail Stop B

299, Los Alamos, New Mexico 87545, USA

ABSTRACT

The Mexico City Air Quality Research Initiative (MARI) is one project that is examining the complex relationship between air pollution, economic growth, societal values, and air quality management policies. The project utilizes a systems approach including computer modeling, comprehensive measurement studies of Mexico City's air pollutants, environmental chemical reaction studies and socioeconomic analyses. MARI, a three-year effort, has three separate components: Task 1 - Modeling and Simulation; Task 2 - Measurement and Characterization; and Task 3 - Strategic Evaluation. In this work we present an overview of the project objectives and progress, highlighting the results obtained during three major field exercises in the Mexico City Basin and some results of simulating control strategies.

INTRODUCTION

Urban air pollution is an environmental problem in many cities around the world which has serious immediate and long-term implications for the health of the population and for the physical environment. Mexico

City, in particular, faces a severe air pollution problem due to a combination of circumstances. The city is in a high mountain basin at a subtropical latitude. The basin setting inhibits dispersion of pollution and contributes to frequent wintertime thermal inversions which further trap pollutants near the surface. The elevation and latitude combine to provide plentiful sunshine which drives atmospheric photochemistry to produce secondary pollutants such as ozone. The 1990 census (XI Censo General de Poblacion y Vivienda de 1990) records that slightly over 15 million people live in the MCMA. There are numerous other nearby communities which are in the airshed region of Mexico City, but which are not included in the definition and population of the MCMA.

More than 30,000 industrial establishments are located in the

600 Air Pollution

MCMA of which 1,500 to 1,800 fall in the medium and large categories. There are 12,000 commercial/service facilities utilizing combustion processes and a large number of non-combustion sources such as dry cleaning, printing and solvent use. The transportation sector includes 2.6 million private vehicles, 56,500 taxicabs, 7,500 buses, 54,500 microbuses, 196,000 gasoline fueled trucks, 60,000 diesel fueled trucks and railway and airport facilities. All of this activity requires fuel: 20 million liters of gasoline and diesel, 1.8 million liters of fuel-oil and 10 million cubic meters of natural gas are used each day [1],

Solving an air pollution problem requires much more than engineering solutions. The Mexico City Air Quality Research Initiative (MARI) utilizes a systems approach including computer modeling, comprehensive measurement studies of Mexico City's air pollutants, environmental chemical reaction studies and socioeconomic analysis. It will provide a set of decision analysis tools to assist

Mexican policy makers in determining optimum strategies from amongst a vast array of options to defeat the air pollution problem. Environmental analysis will be based on models that provide a three-dimensional real-time picture of the atmosphere over Mexico City, including wind flow and turbulence as well as the concentrations of all environmentally important chemical species.

Sponsors for the initiative are Mexico's Petroleos Mexicanos (Pemex) and the U.S. Department of Energy (DOE). Project leadership is provided by the Institute Mexicano del Petroleo (IMP) and the Los Alamos National Laboratory (LAND. The three year collaborative project was signed in July 1990. While the Mexican Petroleum Institute and Los Alamos National Laboratory are designated as the lead institutions for this project, there has been substantial and important support and participation from other institutions in both nations. These include, in Mexico, the Mexican Secretariat for Social Development (Sedesol), formerly Sedue, the Mexican Federal District (DDF), the Mexican National University (UN AM) and the National Polytechnic Institute (IPN). In the U.S., the University of Denver, Carnegie-Mellon University, the University of Utah, the National Center for Atmospheric Research (NCAR), the National Oceanic and Atmospheric Administration (NOAA) and the EPA have all contributed.

MODELING AND SIMULATION

Developing a simulation capability for use in making air quality management decisions requires a system of models. For the most part existing models are being used, as the HOTMAC and RAPTAD codes developed originally at LANL, but with significant effort to adapt them to describe the atmospheric physics and chemistry occurring in the Mexico City Basin. Development of the modeling system is described in more detail in references 2-4. Adaptation and testing of the models will continue based upon satellite remote sensing data, updated emissions inventory, and further assimilation of data from the experimental campaigns. Model results to date, however,

Air Pollution 601

demonstrate good correspondence with the patterns and trends of meteorology and air quality in Mexico City.

Photochemistry is being evaluated and simulated in two different models, an EPA standard photochemical box model, OZIPM-4, and a full three dimensional airshed model (the most sophisticated tool available for air quality management studies). The airshed model being used is the CIT model, developed first at the California Institute of Technology with continuing development at Carnegie-Mellon University.

Airshed simulations of Mexico City air quality were performed at Carnegie-Mellon University, in collaboration with IMP researchers and the codes were then transferred to IMP. The first ever 3-D airshed model simulation done in Mexico City were carried out at IMP in April 1992. This effort is attempting to describe the ozone levels measured across the city on Feb. 22, 1991, a date in the midst of an intensive experimental campaign involving up to 100 U.S. and Mexican scientists. The simulation performed made use of an improved windfield set calculated at LANL and a revised emission inventory. The results are encouraging, but demonstrate the need for updating and improvement in the spatial, temporal and total components of the emissions inventory.

CHARACTERIZATION AND MEASUREMENT

Three field campaigns have been staged in Mexico City as part of this project for the purpose of obtaining comprehensive data sets describing the meteorology, dynamics and chemistry of the Mexico City airshed. Some of the data is used to develop realistic input descriptions for the simulation models, but much of the data is used for comparison to model predictions so that it may be determined if they are accurately portraying Mexico City. This, of course, is essential before the results of mitigation options may be computed. In September, 1990 a meteorological team from Los Alamos and the National Oceanic and Atmospheric Administration conducted tethersonde and ozonesonde measurements for two weeks. Vertical profiles of wind velocity, temperature, humidity and ozone were obtained at each site at up to 1 km above the surface. To our knowledge, this was the first time in Mexico City that a chemical species or pollutant had been measured above the surface. One of the most interesting findings was the existence and persistence throughout the night of an elevated layer containing a high concentration of ozone.

In February, 1991 a major field campaign, and the largest ever in Mexico City for environmental purposes, was conducted over three weeks at several city locations. About 14 different measurement techniques were applied to observe the Mexican atmosphere during that period. They ranged from airborne experiments as instrumented aircrafts and balloons, to state of the art, remote sensing, lidars (Light Detection And Ranging) and infrared detectors, as well as more traditional monitors for meteorological, particle and contaminants

602 Air Pollution observations. More than a dozen participating institutions with almost one hundred people collaborated.

In the first week the tethersonde and lidar experiments were co-located at the Valle de Mexico thermoelectric plant. In the second week the tethersonde and other experiments were moved to a sports stadium at the National Polytechnic Institute and the two lidars, from LANL and the Mexican Institute de Investigaciones Electricas (HE) were set up at the Cinvestav site of the IPN about 1 km to the northeast. In the third week the tethersonde team returned to the Xochimilco area while the HE SOz lidar was moved to the 18 de Marzo refinery and the elastic scattering lidar was moved to the UNAM Botanical Gardens. Concurrent with these efforts the NCAR King Air research aircraft was flying 40 hours of measurement time and the University of Denver automotive emissions remote sensing FEAT experiment was being deployed at several locations around the city.

In March, 1992 a smaller measurements campaign was conducted by LANL, the IMP and the U.S. EPA. Part of the campaign focused on ambient hydrocarbon measurements; information which is vital to the photochemical modeling effort. A portable photoionization detector was used by LANL to measure ambient total hydrocarbon concentrations and to detect time varying concentration trends. The data is somewhat difficult to interpret, but it confirms high ambient concentrations and establishes that there is a distinct concentration increase from 7 a.m. to 10 a.m. [5]. The U.S. EPA, with assistance and support from Sedesol, DDF, and IMP, conducted a canister whole air sampling campaign throughout the month of March at several different city locations. The canisters were returned to EPA at Research Triangle Park and analyzed for CO, CH4 total non methane organic compounds

(NMOC) and speciated hydrocarbons. Six a.m. to 9 a.m. measurements at a northern industrial site (samples on 14 different days) and downtown (samples on 15 different days) showed mean NMOC concentrations of 4.5 and 3.8 ppmC respectively [6]. High hydrocarbon levels were measured city wide which indicates that the automotive fleet is a major contributor, but the highest concentrations were noted in an industrial region of the city, pointing to significant contributions by industry. A summary of the total NMOC concentrations is given in Table 1.

The other focus of the March, 1992 campaign was to measure and gain better understanding of the wind flows at the lower end of the mountain slopes. For this purpose a new miniaturized elastic scattering lidar developed by LANL was deployed at the Rancho del Charro in the western area of the City. The purpose was again to measure dynamics of the aerosol layers and also to use correlation techniques to measure wind velocities in three dimensions. A Doppler sodar (acoustic radar) was also used at the site to obtain vertical profiles of wind velocities for use in interpreting the lidar data.

Air Pollution 603

Table 1. Total non methane organic compounds concentrations obtained during the March 1992, air sampling campaign. Samples were taken between 6 and 9 in the morning (AM) or 12 and 3 in the afternoon (PM).

SITE # of samples Concentration [ppmC]

Pachuca Hgo. 5 0.,243 0.917 0.646 Cuautitlan 5 0,.956 1.693 1.148 1.957 7,.170 4.494 Xalostoc 14 Tlalnepantla (AM) 12 1.549 4.676 3.465 1.559 Tlalnepantla (PM) 5 1.328 1.931 La Merced 15 2.,432 6.184 3.772 Pedregal (AM) 5 1.678 2.565 1.967 Pedregal (PM) 7 0.479 2.358 1.204

Much of the data from the February, 1991 campaign has been analyzed and incorporated into the simulation effort the strategic analysis effort. The campaign was largely directed toward the measurement of the diurnal cycle of atmospheric dynamics with the meteorological instrumentation and the elastic scattering lidar [7-91. As a general result for February, it can be said that during the early morning hours, a concentrated aerosol layer is often present at an altitude of a few hundred meters. After sunrise, and as the level of activity in the city increases, the layer moves upwards, reaching a height well above 1 km by 12-1 p.m.

Other efforts included the measurement and analysis of incident solar radiation, both direct and scattered, and the automotive emissions measurements. The solar radiation analysis is important to the photochemical simulation effort [10] and the automotive emissions are important for the emissions database and for development of mitigation options. The FEAT (Fuel Efficiency Automotive Test) utilizes an infrared source on one side of a lane of traffic and four IR detectors housed together on the other side of the traffic lane to make emissions measurements in real time under actual traffic conditions. The detectors are for sensing CO, COz and hydrocarbons with the fourth channel being a reference. A video camera is triggered by the automobile passing the sensor and the license plate of the vehicle is recorded. In two weeks somewhat over 30,000 measurements were made and the analysis [11] shows that the fleet average emissions are in general high when compared to US vehicles, particularly for CO, however for HC they are comparable to the US fleet although slightly higher.

STRATEGIC EVALUATION

There are many options under consideration for improving the air quality of Mexico City. Each option has costs (of several different kinds), certain projected environmental benefits and then a wide

604 Air Pollution array of less easily quantified social-political-institutional considerations. The Initiative is developing and demonstrating two decision analysis techniques which will provide support for rational decision making. The first is the use of linear programming to select a set of options which meet some objective. The options each have an air quality benefit vis-a-vis one or more pollutants and certain costs including implementation costs and annual operating costs. The cost factor being used in the linear programming model at present is the total capital cost plus five years of operating costs. Cost and benefit information is being gathered on about seventy options.

When certain strategies, or combinations of options, have been formulated by use of the linear programming model, multi-attribute decision theory will be used to evaluate the social-political- institutional factors in combination with environmental, economic and technical factors. This decision theory is a technique to achieve figures of merit for an option given objectives, weights and utility functions. Representatives of IMP, Sedesol, Pemex, DDF, and the State of Mexico met frequently for nearly a year to develop, refine and test the decision tree [12].

RESULTS

The potential of the modeling tools can be assessed by examination of the sensitivity and scenario tests already performed [13,141. The CIT grid model has been used to simulate the formation and evolution of ozone and other species, under a preselected day conditions; it can simulate conditions in three spatial (3D) and one temporal dimension. The grid covers a region 80 km x 70 km x 2 km, enclosing the MCMA. Individual horizontal grid cells are 5 km x 5 km, and the vertical dimension can be divided into as many as 7 vertical layers. The wind fields driving the transport and diffusion of pollution, used as input ' to CIT, is the output generated by the (3D HOTMAC) meteorological model, adapted and calibrated to the Mexico City basin conditions for the preselected day.

As an example of the results, Figure 1 shows a comparison of ozone evolution, for one of the grid cells, of the 3D results obtained by simulating the shutdown of the "18 de Marzo" refinery and the February 19, 1991, conditions. This date has been chosen as the ozone episode base case, due to the amount of data gathered for it; from usual monitoring data plus the intensive monitoring campaign fielded during February 1991. The selected date, besides, is representative of a typical winter day pattern behavior.

The simulations, for the base case and a case with the refinery stacks shut down, outline the influence of altering a point source and show the effects in different city areas. The larger influence found is at the site shown in the figure; this is located West of the refinery. The difference in the ozone maxima is about 22 per cent; yet, in the receptor site located in the SW of the city, the predicted variation was only 5 per cent, as can be seen in Figure 2.

Air Pollution 605

(03) ppm

0.3

Figure 1: Ozone airshed simulation, Acatlan station.

[O3I ppm

-*- 8mm* Cmm* 0.3 O R««. Shutdown

0.2 •

0.1

8 12 16 20 24 Local Time Figure 2: Ozone airshed simulation, Pedregal station.

[O3l ppm

0.2 -

Figure 3: Ozone airshed simulation, La Merced station.

606 Air Pollution

Similar comparisons made for other grid points, show very little, if any, differences, as in Figure 3 where the downtown case is illustrated, together with the monitored data for that site.

To test the capability of simulating combined options, which alter emissions from stationary and mobile sources, two different control strategies were analyzed. Strategy one is a collection of 8 control measures that reduce, for example, total CO emissions 14 per cent and 6 per cent total hydrocarbons. Strategy II includes 13 options that reduce emissions further, 59 per cent CO and 26 percent total reduction of hydrocarbons. Figures 4 and 5 illustrate the results for different grid cells, showing how local conditions affect the ozone behavior. In Fig. 4, Strategy II, has a higher impact in ozone reduction, as expected. However, in Fig. 5, it can be noted that Strategy I instead of decreasing the ozone maxima as expected, it increases it. Strategy II seems to work as desired.

CONCLUSIONS

The components of an air quality management program include the following:

1) Adopt air quality standards 2) Monitor 3) Identify/measure emissions 4) Determine needed reductions and means to accomplish 5) Adopt control measures 6) Implement enforcement 7) Evaluation

In Mexico City air quality standards have been adopted and air quality monitoring is underway. A baseline version of the emission database has been specified but, in common with experience elsewhere, updating and refinement continue. As experience in the South Coast Air Basin of California has shown, this is a difficult task with the emissions database seemingly always underestimated. MARI is primarily concerned with item 4 so that the decisions to adopt control measures (item 5) may be made with sound supporting facts, Obviously, some control measures, such as catalytic converters on automobiles, have already been adopted. The evaluation tools developed in this Initiative will be even more useful when less obvious measures which may each reduce emissions only a small amount have to be considered.

MARI is being cited as a model for international cooperative projects in the area of energy and environmental practices and technologies. A few of the key elements include researchers from both nations working side by side as peers, both nations investing resources and having an interest in the outcome of the project, and the objective being, not advocacy, but application of science to problem solving. In the longer term, financial commitment is needed to maintain and upgrade the capabilities developed within the project.

607 Air Pollution

1031 ppm

0 8tr«»ty I O Mrai*ty II

o S 5 5 3 5 5

LocaJ Tim* Figure 4: Impact of strategies I and II In Cerro de la Estrella station.

IO3I ppm

Local Time Figure 5: Impact of strategies I and II in Acatlan station.

608 Air Pollution

REFERENCES

1. Departamento del Distrito Federal, "Programa Integral Contra la Contaminaci6n Atomosferica", October, 1990.

2. M.D. Williams, G. Sosa, G.E. Streit, X. Cruz, M. Ruiz, A.G. Russell and L.A. McNair, "Development and Testing of an Air Quality Model for Mexico City", LA-UR-92-808, March, 1992.

3. M.E. Ruiz-Santoyo and X. Cruz-Nunez. "Modelos Fotoquimicos de Contaminantes AtmosfeYicos Urbanos". CIENCIA (Revista de la Academia de la Investigaci6n Cientifica). 42, (1991) 99-109.

4. X. Cruz, S. Chavez and M. E. Ruiz. "Analisis de sensibilidad local para un modelo de calidad del aire". Revista de la Sociedad Quimica de Mexico, 36, 1992, 105.

5. G.E. Streit, "Real Time Total Hydrocarbon Measurements in Mexico City", LA-UR-92-1840, June, 1992.

6. Bob Seila, U.S. EPA, private communication.

7. W.M. Porch', W.E. Clements and J.A. Herwehe, "Analysis of Tethered Balloon-borne Measurement in Mexico City September 1990 and February 1991", LA-UR-92-1295, April, 1992.

8. W.M. Porch, L. Auer, W. Clements and P. Mutschlecner, "Air Pollution Mixing Height Studies in Mexico City", LA-UR-92-1296, April, 1992.

9. C.R. Quick, Jr., F.L. Archuleta, D.E. Hof, R.R.. Karl, Jr., J.J. Tiee, W.E. Eichinger, D.B. Holtkamp and L.L. Tellier, "Preliminary Report of the Mexico City 1991 Lidar Measurements Campaign", September, 1992.

10. J.C. Ruiz S., L.G. Ruiz S., S. Eidels-Dubovoi and A Muhlia, "Photolytic Rates for NOz, 03, and HCHO in the Atmosphere of Mexico City", Atmospheric Environment (accepted).

11. S.P. Beaton, G.A. Bishop and D.H. Stedman, "Emission Characteristics of Mexico City Vehicles", J. Air and Waste Management Assoc. (accepted).

12. A.S. Barrera Roldan, A. Corona Juarez, R.W. Hardie, L.R. Mollinedo and G. Thayer, "Metodologia de Atributos Multiples para la

Evaluaci6n de Estrategias", Institute Mexicano del Petroleo, April, 1992.

13. G.E. Streit. "Mexico City Air Quality: Progress of an International Collaborative Project to Define Air Quality Management Options". Supercities International Conference on Environmental

Air Pollution 609

Quality and Sustainable Development, San Francisco, CA. LA-UR-92-2919

October, 1992.

14. F. Guzman, M.E. Ruiz and G. Sosa, "Mexico City Air Quality Research Initiative". World Energy Council, 15th Congress, Vol 1.1, 321, Madrid, September, 1992.